Elon Musk says SpaceX could have a Fable or GPT-6 level AI model within months

Elon Musk is not known for underselling timelines, but his latest prediction for SpaceX’s AI ambitions is striking even by his standards. On September 24, Musk said SpaceX could have a model comparable to Anthropic’s Fable or OpenAI’s GPT-6 ready within two to three months.

What SpaceX is actually building

The AI push at SpaceX is not purely theoretical. The company has already reassigned a few dozen engineers from its Starlink and Starship programs specifically to work on AI development, a meaningful signal of internal prioritization given how resource-constrained those flagship programs tend to be.

SpaceX’s AI work builds on the Grok family of models developed under xAI. Grok 4.5, which launched in June 2026, runs on a 1.5 trillion-parameter base model and entered private beta at SpaceX and Tesla shortly after launch. The model also incorporates data from Cursor, the AI coding tool.

Speaking of Cursor: SpaceX acquired the startup, with reported deal values ranging between $6 billion and $60 billion in stock. Musk also outlined an aggressive release cadence. SpaceX intends to ship new foundation models every month through the rest of 2026.

Where this fits in the broader AI race

Musk went further than the two-to-three month window. He projected that if SpaceX maintains its current growth trajectory, the company could take a leading position in AI within six months.

What to watch and why it matters

The monthly release strategy is perhaps the most consequential detail. If SpaceX follows through, it introduces a new kind of competitive pressure: not just better models, but faster model improvement cycles.

There is also a talent dimension worth tracking. SpaceX pulling engineers from Starlink and Starship is not cost-free. Both programs have their own ambitious timelines, and the reallocation creates a resource tension that will only grow if the AI push intensifies.